CPU Performance Monitor. More...

#include <cpucounters.h>

Classes
struct	CustomCoreEventDescription
	Custom Core event description. More...

struct	CustomIIOEventDescription

struct	ExtendedCustomCoreEventDescription
	Extended custom core event description. More...

struct	RawPMUConfig

struct	SimplePCIeDevInfo

Public Types
enum	{ MAX_C_STATE = 10 }

enum	ProgramMode { DEFAULT_EVENTS = 0, CUSTOM_CORE_EVENTS = 1, EXT_CUSTOM_CORE_EVENTS = 2, INVALID_MODE }
	Mode of programming (parameter in the program() method) More...

enum	ErrorCode { Success = 0, MSRAccessDenied = 1, PMUBusy = 2, UnknownError }
	Return codes (e.g. for program(..) method)

enum	PerfmonField { INVALID, OPCODE, EVENT_SELECT, UMASK, RESET, EDGE_DET, IGNORED, OVERFLOW_ENABLE, ENABLE, INVERT, THRESH, CH_MASK, FC_MASK, H_EVENT_NAME, V_EVENT_NAME, MULTIPLIER, DIVIDER, COUNTER_INDEX }

enum	PCIeWidthMode { X1, X4, X8, X16, XFF }

enum	{ IIO_CBDMA = 0, IIO_PCIe0 = 1, IIO_PCIe1 = 2, IIO_PCIe2 = 3, IIO_MCP0 = 4, IIO_MCP1 = 5, IIO_STACK_COUNT = 6 }

enum	SkylakeIIOStacks { SKX_IIO_CBDMA_DMI = 0, SKX_IIO_PCIe0 = 1, SKX_IIO_PCIe1 = 2, SKX_IIO_PCIe2 = 3, SKX_IIO_MCP0 = 4, SKX_IIO_MCP1 = 5, SKX_IIO_STACK_COUNT = 6 }

enum	IcelakeIIOStacks { ICX_IIO_PCIe0 = 0, ICX_IIO_PCIe1 = 1, ICX_IIO_MCP0 = 2, ICX_IIO_PCIe2 = 3, ICX_IIO_PCIe3 = 4, ICX_IIO_CBDMA_DMI = 5, ICX_IIO_STACK_COUNT = 6 }

enum	SnowridgeIIOStacks { SNR_IIO_QAT = 0, SNR_IIO_CBDMA_DMI = 1, SNR_IIO_NIS = 2, SNR_IIO_HQM = 3, SNR_IIO_PCIe0 = 4, SNR_IIO_STACK_COUNT = 5 }

enum	MSREventPosition { index = 0, type = 1 }

enum	MSRType { Static = 0, Freerun = 1 }

enum	EventPosition { TOR_OCCUPANCY = 0, TOR_INSERTS = 1, REQUESTS_ALL = 2, REQUESTS_LOCAL = 3 }

enum	{ OCR0Pos = 1, OCR1Pos = 2, LoadLatencyPos = 3, FrontendPos = 4 }

enum	SupportedCPUModels { NEHALEM_EP = 26, NEHALEM = 30, ATOM = 28, ATOM_2 = 53, CENTERTON = 54, BAYTRAIL = 55, AVOTON = 77, CHERRYTRAIL = 76, APOLLO_LAKE = 92, DENVERTON = 95, SNOWRIDGE = 134, CLARKDALE = 37, WESTMERE_EP = 44, NEHALEM_EX = 46, WESTMERE_EX = 47, SANDY_BRIDGE = 42, JAKETOWN = 45, IVY_BRIDGE = 58, HASWELL = 60, HASWELL_ULT = 69, HASWELL_2 = 70, IVYTOWN = 62, HASWELLX = 63, BROADWELL = 61, BROADWELL_XEON_E3 = 71, BDX_DE = 86, SKL_UY = 78, KBL = 158, KBL_1 = 142, CML = 166, CML_1 = 165, ICL = 126, ICL_1 = 125, RKL = 167, TGL = 140, TGL_1 = 141, ADL = 151, ADL_1 = 154, BDX = 79, KNL = 87, SKL = 94, SKX = 85, ICX_D = 108, ICX = 106, END_OF_MODEL_LIST = 0x0ffff }
	Identifiers of supported CPU models.

enum	PCIeEventCode { PCIeRdCur = 0x19E, PCIeNSRd = 0x1E4, PCIeWiLF = 0x194, PCIeItoM = 0x19C, PCIeNSWr = 0x1E5, PCIeNSWrF = 0x1E6, RFO = 0x180, CRd = 0x181, DRd = 0x182, PRd = 0x187, WiL = 0x18F, ItoM = 0x1C8, SKX_RFO = 0x200, SKX_CRd = 0x201, SKX_DRd = 0x202, SKX_PRd = 0x207, SKX_WiL = 0x20F, SKX_RdCur = 0x21E, SKX_ItoM = 0x248 }

enum	ChaPipelineQueue { None, IRQ, PRQ }

enum	CBoEventTid { RFOtid = 0x3E, ItoMtid = 0x3E }

typedef std::pair< std::array< uint64, 5 >, std::string >	RawEventConfig

typedef std::map< std::string, RawPMUConfig >	RawPMUConfigs

Public Member Functions
bool	isCoreCStateResidencySupported (int state)
	Returns true if the specified core C-state residency metric is supported.

bool	isPackageCStateResidencySupported (int state)
	Returns true if the specified package C-state residency metric is supported.

void	restoreOutput ()
	Restores output, closes output file if opened.

void	setRunState (int new_state)
	Set Run State.

int	getRunState (void)
	Returns program's Run State.

bool	isBlocked (void)

void	setBlocked (const bool new_blocked)

bool	isHWTMAL1Supported () const
	check if TMA level 1 metrics are supported

bool	isSecureBoot () const
	check if in secure boot mode

bool	useLinuxPerfForUncore () const
	true if Linux perf for uncore PMU programming should AND can be used internally

SystemRoot const &	getSystemTopology () const
	The system, sockets, uncores, cores and threads are structured like a tree. More...

void	printDetailedSystemTopology ()
	prints detailed system topology

bool	QOSMetricAvailable () const
	checks if QOS monitoring support present More...

bool	L3QOSMetricAvailable () const
	checks L3 cache support for QOS present More...

bool	L3CacheOccupancyMetricAvailable () const
	checks if L3 cache monitoring present More...

bool	CoreLocalMemoryBWMetricAvailable () const
	checks if local memory bandwidth monitoring present More...

bool	CoreRemoteMemoryBWMetricAvailable () const
	checks if total memory bandwidth monitoring present More...

unsigned	getMaxRMID () const
	returns the max number of RMID supported by socket More...

uint32	getMaxNumOfCBoxes () const
	Returns the number of CBO or CHA units per socket.

uint32	getMaxNumOfIIOStacks () const
	Returns the number of IIO stacks per socket.

bool	good ()
	Checks the status of PCM object. More...

const std::string &	getErrorMessage () const
	Returns the error message. More...

ErrorCode	program (const ProgramMode mode_=DEFAULT_EVENTS, const void *parameter_=NULL, const bool silent=false, const int pid=-1)
	Programs performance counters. More...

void	checkError (const ErrorCode code)
	checks the error and suggests solution and/or exits the process More...

ErrorCode	programServerUncoreLatencyMetrics (bool enable_pmm)
	Programs uncore latency counters on microarchitectures codename SandyBridge-EP and later Xeon uarch. More...

ErrorCode	programServerUncorePowerMetrics (int mc_profile, int pcu_profile, int *freq_bands=NULL)
	Programs uncore power/energy counters on microarchitectures codename SandyBridge-EP and later Xeon uarch. More...

ErrorCode	programServerUncoreMemoryMetrics (const ServerUncoreMemoryMetrics &metrics, int rankA=-1, int rankB=-1)

ErrorCode	program (const RawPMUConfigs &curPMUConfigs, const bool silent=false, const int pid=-1)

std::pair< unsigned, unsigned >	getOCREventNr (const int event, const unsigned coreID) const

void	freezeServerUncoreCounters ()
	Freezes uncore event counting (works only on microarchitecture codename SandyBridge-EP and IvyTown)

void	unfreezeServerUncoreCounters ()
	Unfreezes uncore event counting (works only on microarchitecture codename SandyBridge-EP and IvyTown)

ServerUncoreCounterState	getServerUncoreCounterState (uint32 socket)
	Reads the power/energy counter state of a socket (works only on microarchitecture codename SandyBridge-EP) More...

void	cleanup (const bool silent=false)
	Cleanups resources and stops performance counting. More...

void	resetPMU ()
	Forces PMU reset. More...

void	getAllCounterStates (SystemCounterState &systemState, std::vector< SocketCounterState > &socketStates, std::vector< CoreCounterState > &coreStates, const bool readAndAggregateSocketUncoreCounters=true)
	Reads all counter states (including system, sockets and cores) More...

void	getUncoreCounterStates (SystemCounterState &systemState, std::vector< SocketCounterState > &socketStates)
	Reads uncore counter states (including system and sockets) but no core counters. More...

bool	isCoreOnline (int32 os_core_id) const
	Return true if the core in online. More...

bool	isSocketOnline (int32 socket_id) const
	Return true if the socket in online. More...

SystemCounterState	getSystemCounterState ()
	Reads the counter state of the system. More...

SocketCounterState	getSocketCounterState (uint32 socket)
	Reads the counter state of a socket. More...

CoreCounterState	getCoreCounterState (uint32 core)
	Reads the counter state of a (logical) core. More...

uint32	getNumCores () const
	Reads number of logical cores in the system. More...

uint32	getNumOnlineCores () const
	Reads number of online logical cores in the system. More...

uint32	getNumSockets () const
	Reads number of sockets (CPUs) in the system. More...

uint32	getNumOnlineSockets () const
	Reads number of online sockets (CPUs) in the system. More...

uint32	getThreadsPerCore () const
	Reads how many hardware threads has a physical core "Hardware thread" is a logical core in a different terminology. If Intel(r) Hyperthreading(tm) is enabled then this function returns 2. More...

bool	getSMT () const
	Checks if SMT (HyperThreading) is enabled. More...

uint64	getNominalFrequency () const
	Reads the nominal core frequency. More...

uint32	getL3ScalingFactor () const
	runs CPUID.0xF.0x01 to get the L3 up scaling factor to calculate L3 Occupancy Scaling factor is returned in EBX register after running the CPU instruction More...

bool	isSomeCoreOfflined ()
	runs CPUID.0xB.0x01 to get maximum logical cores (including SMT) per socket. max_lcores_per_socket is returned in EBX[15:0]. Compare this value with number of cores per socket detected in the system to see if some cores are offlined More...

int32	getMaxCustomCoreEvents ()
	Returns the maximum number of custom (general-purpose) core events supported by CPU.

uint32	getCPUModel () const
	Reads CPU model id. More...

uint32	getCPUStepping () const
	Reads CPU stepping id. More...

int32	getThreadId (uint32 os_id) const
	Determines physical thread of given processor ID within a core. More...

int32	getCoreId (uint32 os_id) const
	Determines physical core of given processor ID within a socket. More...

int32	getTileId (uint32 os_id) const
	Determines physical tile (cores sharing L2 cache) of given processor ID. More...

int32	getSocketId (uint32 core_id) const
	Determines socket of given core. More...

uint64	getQPILinksPerSocket () const
	Returns the number of Intel(r) Quick Path Interconnect(tm) links per socket. More...

uint32	getMCPerSocket () const
	Returns the number of detected integrated memory controllers per socket.

size_t	getMCChannelsPerSocket () const
	Returns the total number of detected memory channels on all integrated memory controllers per socket.

size_t	getMCChannels (uint32 socket, uint32 controller) const
	Returns the number of detected memory channels on given integrated memory controllers. More...

size_t	getEDCChannelsPerSocket () const
	Returns the total number of detected memory channels on all integrated memory controllers per socket.

uint32	getMaxIPC () const
	Returns the max number of instructions per cycle. More...

uint64	getPCUFrequency () const
	Returns the frequency of Power Control Unit.

bool	isServerCPU () const
	Returns whether it is a server part.

bool	isClientCPU () const
	Returns whether it is a client part.

uint64	getTickCount (uint64 multiplier=1000, uint32 core=0)
	Return TSC timer value in time units. More...

uint64	getInvariantTSC_Fast (uint32 core=0)

uint64	getUncoreClocks (const uint32 socket_)
	Returns uncore clock ticks on specified socket.

uint64	getQPILinkSpeed (uint32 socketNr, uint32 linkNr) const
	Return QPI Link Speed in GBytes/second. More...

double	getJoulesPerEnergyUnit () const
	Returns how many joules are in an internal processor energy unit.

int32	getPackageThermalSpecPower () const
	Returns thermal specification power of the package domain in Watt.

int32	getPackageMinimumPower () const
	Returns minimum power derived from electrical spec of the package domain in Watt.

int32	getPackageMaximumPower () const
	Returns maximum power derived from electrical spec of the package domain in Watt.

void	disableJKTWorkaround ()

void	programPCIeEventGroup (eventGroup_t &eventGroup)
	Program uncore PCIe monitoring event(s) More...

uint64	getPCIeCounterData (const uint32 socket_, const uint32 ctr_)

void	programCbo (const uint64 *events, const uint32 opCode=0, const uint32 nc_=0, const uint32 llc_lookup_tid_filter=0, const uint32 loc=1, const uint32 rem=1)
	Program CBO (or CHA on SKX+) counters. More...

void	programCboRaw (const uint64 *events, const uint64 filter0, const uint64 filter1)
	Program CBO (or CHA on SKX+) counters. More...

PCIeCounterState	getPCIeCounterState (const uint32 socket_, const uint32 ctr_=0)
	Get the state of PCIe counter(s) More...

void	programIIOCounters (uint64 rawEvents[4], int IIOStack=-1)
	Program uncore IIO events. More...

void	programIRPCounters (uint64 rawEvents[4], int IIOStack=-1)
	Program uncore IRP events. More...

IIOCounterState	getIIOCounterState (int socket, int IIOStack, int counter)
	Get the state of IIO counter. More...

void	getIIOCounterStates (int socket, int IIOStack, IIOCounterState *result)
	Get the states of the four IIO counters in bulk (faster than four single reads) More...

uint64	extractCoreGenCounterValue (uint64 val)

uint64	extractCoreFixedCounterValue (uint64 val)

uint64	extractUncoreGenCounterValue (uint64 val)

uint64	extractUncoreFixedCounterValue (uint64 val)

uint64	extractQOSMonitoring (uint64 val)

const char *	getUArchCodename (const int32 cpu_model_=-1) const
	Get a string describing the codename of the processor microarchitecture. More...

std::string	getCPUFamilyModelString ()

void	enableForceRTMAbortMode (const bool silent=false)
	Enables "force all RTM transaction abort" mode also enabling 4+ programmable counters on Skylake generation processors.

bool	isForceRTMAbortModeEnabled () const
	queries status of "force all RTM transaction abort" mode

void	disableForceRTMAbortMode (const bool silent=false)
	Disables "force all RTM transaction abort" mode restricting the number of programmable counters on Skylake generation processors to 3.

bool	isForceRTMAbortModeAvailable () const
	queries availability of "force all RTM transaction abort" mode

int64	getCPUMicrocodeLevel () const
	Get microcode level (returns -1 if retrieval not supported due to some restrictions)

bool	isAtom () const
	returns true if CPU is Atom-based

bool	packageEnergyMetricsAvailable () const

bool	dramEnergyMetricsAvailable () const

bool	packageThermalMetricsAvailable () const

bool	outgoingQPITrafficMetricsAvailable () const

bool	incomingQPITrafficMetricsAvailable () const

bool	localMemoryRequestRatioMetricAvailable () const

bool	qpiUtilizationMetricsAvailable () const

bool	memoryTrafficMetricsAvailable () const

bool	MCDRAMmemoryTrafficMetricsAvailable () const

bool	memoryIOTrafficMetricAvailable () const

bool	IIOEventsAvailable () const

bool	uncoreFrequencyMetricAvailable () const

bool	LatencyMetricsAvailable () const

bool	DDRLatencyMetricsAvailable () const

bool	PMMTrafficMetricsAvailable () const

bool	LLCReadMissLatencyMetricsAvailable () const

bool	hasBecktonUncore () const

bool	hasPCICFGUncore () const

bool	isSkxCompatible () const

bool	hasUPI () const

const char *	xPI () const

bool	hasCHA () const

bool	supportsHLE () const

bool	supportsRTM () const

bool	supportsRDTSCP () const

bool	useSkylakeEvents () const

bool	hasClientMCCounters () const

double	getBytesPerFlit () const

double	getDataBytesPerFlit () const

double	getBytesPerLinkCycle () const

double	getBytesPerLinkTransfer () const

void	setupCustomCoreEventsForNuma (PCM::ExtendedCustomCoreEventDescription &conf) const
	Setup ExtendedCustomCoreEventDescription object to read offcore (numa) counters for each processor type. More...

bool	isActiveRelativeFrequencyAvailable () const

Static Public Member Functions
static void	setOutput (const std::string filename, const bool cerrToo=false)
	Redirects output destination to provided file, instead of std::cout and std::cerr (optional)

static PCM *	getInstance ()
	Returns PCM object. More...

static int	getCPUModelFromCPUID ()
	Returns cpu model id number from cpuid instruction.

static bool	initWinRing0Lib ()
	Loads and initializes Winring0 third party library for access to processor model specific and PCI configuration registers. More...

static std::string	getCPUBrandString ()
	Get Brand string of processor.

static bool	isAtom (const int32 cpu_model_)
	returns true if CPU model is Atom-based

static bool	hasUPI (const int32 cpu_model_)

static double	getBytesPerFlit (int32 cpu_model_)

static double	getDataBytesPerFlit (int32 cpu_model_)

static double	getFlitsPerLinkCycle (int32 cpu_model_)

static double	getBytesPerLinkCycle (int32 cpu_model_)

static double	getLinkTransfersPerLinkCycle ()

Friends
class	BasicCounterState

class	UncoreCounterState

class	Socket

class	ServerUncore

class	PerfVirtualControlRegister

class	Aggregator

class	ServerPCICFGUncore

Detailed Description

CPU Performance Monitor.

This singleton object needs to be instantiated for each process before accessing counting and measuring routines

Member Enumeration Documentation

◆ ProgramMode

enum pcm::PCM::ProgramMode

Mode of programming (parameter in the program() method)

Enumerator
DEFAULT_EVENTS	Default choice of events, the additional parameter is not needed and ignored
CUSTOM_CORE_EVENTS	Custom set of core events specified in the parameter to the program method. The parameter must be a pointer to array of four `CustomCoreEventDescription` values
EXT_CUSTOM_CORE_EVENTS	Custom set of core events specified in the parameter to the program method. The parameter must be a pointer to a `ExtendedCustomCoreEventDescription` data structure
INVALID_MODE	Non-programmed mode

Member Function Documentation

◆ checkError()

void pcm::PCM::checkError ( const ErrorCode code )

checks the error and suggests solution and/or exits the process

Parameters

code	error code from the 'program' call

References resetPMU().

◆ cleanup()

void pcm::PCM::cleanup ( const bool silent = false )

Cleanups resources and stops performance counting.

One needs to call this method when your program finishes or/and you are not going to use the performance counting routines anymore.

References disableForceRTMAbortMode().

◆ CoreLocalMemoryBWMetricAvailable()

bool pcm::PCM::CoreLocalMemoryBWMetricAvailable ( ) const

checks if local memory bandwidth monitoring present

Returns: true or false

References L3QOSMetricAvailable(), and QOSMetricAvailable().

◆ CoreRemoteMemoryBWMetricAvailable()

bool pcm::PCM::CoreRemoteMemoryBWMetricAvailable ( ) const

checks if total memory bandwidth monitoring present

Returns: true or false

References L3QOSMetricAvailable(), and QOSMetricAvailable().

◆ getAllCounterStates()

void pcm::PCM::getAllCounterStates	(	SystemCounterState &	systemState,
		std::vector< SocketCounterState > &	socketStates,
		std::vector< CoreCounterState > &	coreStates,
		const bool	readAndAggregateSocketUncoreCounters = `true`
	)

Reads all counter states (including system, sockets and cores)

Parameters

systemState	system counter state (return parameter)
socketStates	socket counter states (return parameter)
coreStates	core counter states (return parameter)
readAndAggregateSocketUncoreCounters	read and aggregate socket uncore counters

References isCoreOnline().

◆ getCoreCounterState()

CoreCounterState pcm::PCM::getCoreCounterState ( uint32 core )

Reads the counter state of a (logical) core.

Be aware that during the measurement other threads may be scheduled on the same core by the operating system (this is called context-switching). The performance events caused by these threads will be counted as well.

\param core core id
\return State of counters in the core

Referenced by pcm::getCoreCounterState().

◆ getCoreId()

int32 pcm::PCM::getCoreId ( uint32 os_id ) const

inline

Determines physical core of given processor ID within a socket.

Parameters

os_id processor identifier

Returns: physical core identifier

◆ getCPUModel()

uint32 pcm::PCM::getCPUModel ( ) const

inline

Reads CPU model id.

Returns: CPU model ID

Referenced by pcm::getDRAMClocks(), and pcm::getDRAMConsumedJoules().

◆ getCPUStepping()

uint32 pcm::PCM::getCPUStepping ( ) const

inline

Reads CPU stepping id.

Returns: CPU stepping ID

◆ getErrorMessage()

const std::string& pcm::PCM::getErrorMessage ( ) const

inline

Returns the error message.

Call this when good() returns false, otherwise return an empty string

◆ getIIOCounterState()

IIOCounterState pcm::PCM::getIIOCounterState	(	int	socket,
		int	IIOStack,
		int	counter
	)

Get the state of IIO counter.

Parameters

socket	socket of the IIO stack
IIOStack	id of the IIO stack

Returns: State of IIO counter

◆ getIIOCounterStates()

void pcm::PCM::getIIOCounterStates	(	int	socket,
		int	IIOStack,
		IIOCounterState *	result
	)

Get the states of the four IIO counters in bulk (faster than four single reads)

Parameters

socket	socket of the IIO stack
IIOStack	id of the IIO stack
result	states of IIO counters (array of four IIOCounterState elements)

◆ getInstance()

PCM * pcm::PCM::getInstance ( )

static

Returns PCM object.

Returns PCM object. If the PCM has not been created before than an instance is created. PCM is a singleton.

Returns: Pointer to PCM object

Referenced by pcm::ServerPCICFGUncore::computeQPISpeed(), pcm::getActiveAverageFrequency(), pcm::getActiveRelativeFrequency(), pcm::getAllIncomingQPILinkBytes(), pcm::getAllOutgoingQPILinkBytes(), pcm::getAverageUncoreFrequency(), pcm::getBackendBound(), pcm::getBadSpeculation(), pcm::getBytesReadFromEDC(), pcm::getBytesReadFromMC(), pcm::getBytesReadFromPMM(), pcm::getBytesWrittenToEDC(), pcm::getBytesWrittenToMC(), pcm::getBytesWrittenToPMM(), pcm::getConsumedJoules(), pcm::getCoreCounterState(), pcm::getCoreCStateResidency(), pcm::getCoreIPC(), pcm::getDRAMClocks(), pcm::getDRAMConsumedJoules(), pcm::getEDCCounter(), pcm::getFrontendBound(), pcm::getGTRequestBytesFromMC(), pcm::getIARequestBytesFromMC(), pcm::getIncomingQPILinkBytes(), pcm::getIncomingQPILinkUtilization(), pcm::getIORequestBytesFromMC(), pcm::getL2CacheHitRatio(), pcm::getL2CacheHits(), pcm::getL2CacheMisses(), pcm::getL3CacheHitRatio(), pcm::getL3CacheHits(), pcm::getL3CacheHitsNoSnoop(), pcm::getL3CacheHitsSnoop(), pcm::getL3CacheMisses(), pcm::getL3CacheOccupancy(), pcm::getLLCReadMissLatency(), pcm::getLocalMemoryBW(), pcm::getLocalMemoryRequestRatio(), pcm::getOutgoingQPILinkBytes(), pcm::getOutgoingQPILinkUtilization(), pcm::getPackageCStateResidency(), pcm::getQPItoMCTrafficRatio(), pcm::getRemoteMemoryBW(), pcm::getRetiring(), pcm::getSocketCounterState(), pcm::getSocketIncomingQPILinkBytes(), pcm::getSystemCounterState(), pcm::getTotalExecUsage(), PCMServiceNS::PCMService::OnStart(), pcm::ServerPCICFGUncore::program(), pcm::ServerPCICFGUncore::programServerUncoreMemoryMetrics(), and pcm::ServerPCICFGUncore::reportQPISpeed().

◆ getL3ScalingFactor()

uint32 pcm::PCM::getL3ScalingFactor ( ) const

runs CPUID.0xF.0x01 to get the L3 up scaling factor to calculate L3 Occupancy Scaling factor is returned in EBX register after running the CPU instruction

Returns: L3 up scaling factor

◆ getMaxIPC()

uint32 pcm::PCM::getMaxIPC ( ) const

inline

Returns the max number of instructions per cycle.

Returns: max number of instructions per cycle

◆ getMaxRMID()

unsigned pcm::PCM::getMaxRMID ( ) const

returns the max number of RMID supported by socket

Returns: maximum number of RMID supported by socket

◆ getMCChannels()

size_t pcm::PCM::getMCChannels	(	uint32	socket,
		uint32	controller
	)		const

inline

Returns the number of detected memory channels on given integrated memory controllers.

Parameters

socket	socket
controller	controller

◆ getNominalFrequency()

uint64 pcm::PCM::getNominalFrequency ( ) const

Reads the nominal core frequency.

Returns: Nominal frequency in Hz

Referenced by pcm::getActiveAverageFrequency(), pcm::getIncomingQPILinkUtilization(), pcm::getOutgoingQPILinkBytes(), pcm::getOutgoingQPILinkUtilization(), and getTickCount().

◆ getNumCores()

uint32 pcm::PCM::getNumCores ( ) const

Reads number of logical cores in the system.

Returns: Number of logical cores in the system

Referenced by pcm::getCoreIPC(), and pcm::getTotalExecUsage().

◆ getNumOnlineCores()

uint32 pcm::PCM::getNumOnlineCores ( ) const

Reads number of online logical cores in the system.

Returns: Number of online logical cores in the system

Referenced by pcm::getCoreIPC(), pcm::getIncomingQPILinkUtilization(), pcm::getOutgoingQPILinkBytes(), pcm::getOutgoingQPILinkUtilization(), and pcm::getTotalExecUsage().

◆ getNumOnlineSockets()

uint32 pcm::PCM::getNumOnlineSockets ( ) const

Reads number of online sockets (CPUs) in the system.

Returns: Number of online sockets in the system

◆ getNumSockets()

uint32 pcm::PCM::getNumSockets ( ) const

Reads number of sockets (CPUs) in the system.

Returns: Number of sockets in the system

Referenced by pcm::getAllIncomingQPILinkBytes(), and pcm::getAllOutgoingQPILinkBytes().

◆ getPCIeCounterState()

PCIeCounterState pcm::PCM::getPCIeCounterState	(	const uint32	socket_,
		const uint32	ctr_ = `0`
	)

Get the state of PCIe counter(s)

Parameters

socket_ socket of the PCIe controller

Returns: State of PCIe counter(s)

◆ getQPILinkSpeed()

uint64 pcm::PCM::getQPILinkSpeed	(	uint32	socketNr,
		uint32	linkNr
	)		const

inline

Return QPI Link Speed in GBytes/second.

Warning: Works only for Nehalem-EX (Xeon 7500) and Xeon E7 and E5 processors

Returns: QPI Link Speed in GBytes/second

Referenced by pcm::getIncomingQPILinkUtilization(), pcm::getOutgoingQPILinkBytes(), and pcm::getOutgoingQPILinkUtilization().

◆ getQPILinksPerSocket()

uint64 pcm::PCM::getQPILinksPerSocket ( ) const

inline

Returns the number of Intel(r) Quick Path Interconnect(tm) links per socket.

Returns: number of QPI links per socket

◆ getServerUncoreCounterState()

ServerUncoreCounterState pcm::PCM::getServerUncoreCounterState ( uint32 socket )

Reads the power/energy counter state of a socket (works only on microarchitecture codename SandyBridge-EP)

Parameters

socket socket id

Returns: State of power counters in the socket

◆ getSMT()

bool pcm::PCM::getSMT ( ) const

Checks if SMT (HyperThreading) is enabled.

Returns: true iff SMT (HyperThreading) is enabled.

◆ getSocketCounterState()

SocketCounterState pcm::PCM::getSocketCounterState ( uint32 socket )

Reads the counter state of a socket.

Parameters

socket socket id

Returns: State of counters in the socket

References isCoreOnline().

Referenced by pcm::getSocketCounterState().

◆ getSocketId()

int32 pcm::PCM::getSocketId ( uint32 core_id ) const

inline

Determines socket of given core.

Parameters

core_id core identifier

Returns: socket identifier

◆ getSystemCounterState()

SystemCounterState pcm::PCM::getSystemCounterState ( )

Reads the counter state of the system.

System consists of several sockets (CPUs). Socket has a CPU in it. Socket (CPU) consists of several (logical) cores.

Returns: State of counters in the entire system

References isCoreOnline().

Referenced by pcm::getSystemCounterState().

◆ getSystemTopology()

SystemRoot const& pcm::PCM::getSystemTopology ( ) const

inline

The system, sockets, uncores, cores and threads are structured like a tree.

Returns: a reference to a const System object representing the root of the tree

◆ getThreadId()

int32 pcm::PCM::getThreadId ( uint32 os_id ) const

inline

Determines physical thread of given processor ID within a core.

Parameters

os_id processor identifier

Returns: physical thread identifier

◆ getThreadsPerCore()

uint32 pcm::PCM::getThreadsPerCore ( ) const

Reads how many hardware threads has a physical core "Hardware thread" is a logical core in a different terminology. If Intel(r) Hyperthreading(tm) is enabled then this function returns 2.

Returns: Number of hardware threads per physical core

Referenced by pcm::getCoreIPC(), and pcm::getTotalExecUsage().

◆ getTickCount()

uint64 pcm::PCM::getTickCount	(	uint64	multiplier = `1000`,
		uint32	core = `0`
	)

Return TSC timer value in time units.

Parameters

multiplier	use 1 for seconds, 1000 for ms, 1000000 for mks, etc (default is 1000: ms)
core	core to read on-chip TSC value (default is 0)

Returns: time counter value

References getNominalFrequency().

◆ getTileId()

int32 pcm::PCM::getTileId ( uint32 os_id ) const

inline

Determines physical tile (cores sharing L2 cache) of given processor ID.

Parameters

os_id processor identifier

Returns: physical tile identifier

◆ getUArchCodename()

const char * pcm::PCM::getUArchCodename ( const int32 cpu_model_ = -1 ) const

Get a string describing the codename of the processor microarchitecture.

Parameters

cpu_model_ cpu model (if no parameter provided the codename of the detected CPU is returned)

◆ getUncoreCounterStates()

void pcm::PCM::getUncoreCounterStates	(	SystemCounterState &	systemState,
		std::vector< SocketCounterState > &	socketStates
	)

Reads uncore counter states (including system and sockets) but no core counters.

Parameters

systemState	system counter state (return parameter)
socketStates	socket counter states (return parameter)

◆ good()

bool pcm::PCM::good ( )

Checks the status of PCM object.

Call this method to check if PCM gained access to model specific registers. The method is deprecated, see program error code instead.

Returns: true iff access to model specific registers works without problems

◆ initWinRing0Lib()

static bool pcm::PCM::initWinRing0Lib ( )

static

Loads and initializes Winring0 third party library for access to processor model specific and PCI configuration registers.

Returns: returns true in case of success

Referenced by pcm::Driver::start().

◆ isCoreOnline()

bool pcm::PCM::isCoreOnline ( int32 os_core_id ) const

Return true if the core in online.

Parameters

os_core_id OS core id

Referenced by getAllCounterStates(), getSocketCounterState(), and getSystemCounterState().

◆ isSocketOnline()

bool pcm::PCM::isSocketOnline ( int32 socket_id ) const

Return true if the socket in online.

Parameters

socket_id OS socket id

References getCPUBrandString(), and isAtom().

◆ isSomeCoreOfflined()

bool pcm::PCM::isSomeCoreOfflined ( )

runs CPUID.0xB.0x01 to get maximum logical cores (including SMT) per socket. max_lcores_per_socket is returned in EBX[15:0]. Compare this value with number of cores per socket detected in the system to see if some cores are offlined

Returns: true iff max_lcores_per_socket == number of cores per socket detected

◆ L3CacheOccupancyMetricAvailable()

bool pcm::PCM::L3CacheOccupancyMetricAvailable ( ) const

checks if L3 cache monitoring present

Returns: true or false

References L3QOSMetricAvailable(), and QOSMetricAvailable().

◆ L3QOSMetricAvailable()

bool pcm::PCM::L3QOSMetricAvailable ( ) const

checks L3 cache support for QOS present

Returns: true or false

References isSecureBoot().

Referenced by CoreLocalMemoryBWMetricAvailable(), CoreRemoteMemoryBWMetricAvailable(), and L3CacheOccupancyMetricAvailable().

◆ program()

PCM::ErrorCode pcm::PCM::program	(	const ProgramMode	mode_ = `DEFAULT_EVENTS`,
		const void *	parameter_ = `NULL`,
		const bool	silent = `false`,
		const int	pid = `-1`
	)

Programs performance counters.

Parameters

mode_	mode of programming, see ProgramMode definition
parameter_	optional parameter for some of programming modes
silent	set to true to silence diagnostic messages
pid	restrict core metrics only to specified pid (process id) Call this method before you start using the performance counting routines.

Warning: Using this routines with other tools that program Performance Monitoring Units (PMUs) on CPUs is not recommended because PMU can not be shared. Tools that are known to program PMUs: Intel(r) VTune(tm), Intel(r) Performance Tuning Utility (PTU). This code may make VTune or PTU measurements invalid. VTune or PTU measurement may make measurement with this code invalid. Please enable either usage of these routines or VTune/PTU/etc.

References CUSTOM_CORE_EVENTS, EXT_CUSTOM_CORE_EVENTS, isAtom(), and isHWTMAL1Supported().

◆ programCbo()

void pcm::PCM::programCbo	(	const uint64 *	events,
		const uint32	opCode = `0`,
		const uint32	nc_ = `0`,
		const uint32	llc_lookup_tid_filter = `0`,
		const uint32	loc = `1`,
		const uint32	rem = `1`
	)

Program CBO (or CHA on SKX+) counters.

Parameters

events	array with four raw event values
opCode	opcode match filter
nc_	match non-coherent requests
llc_lookup_tid_filter	filter for LLC lookup event filter and TID filter (core and thread ID)
loc	match on local node target
rem	match on remote node target

◆ programCboRaw()

void pcm::PCM::programCboRaw	(	const uint64 *	events,
		const uint64	filter0,
		const uint64	filter1
	)

Program CBO (or CHA on SKX+) counters.

Parameters

events	array with four raw event values
filter0	raw filter value
filter1	raw filter1 value

◆ programIIOCounters()

void pcm::PCM::programIIOCounters	(	uint64	rawEvents[4],
		int	IIOStack = `-1`
	)

Program uncore IIO events.

Parameters

rawEvents	events to program (raw format)
IIOStack	id of the IIO stack to program (-1 for all, if parameter omitted)

◆ programIRPCounters()

void pcm::PCM::programIRPCounters	(	uint64	rawEvents[4],
		int	IIOStack = `-1`
	)

Program uncore IRP events.

Parameters

rawEvents	events to program (raw format)
IIOStack	id of the IIO stack to program (-1 for all, if parameter omitted)

◆ programPCIeEventGroup()

void pcm::PCM::programPCIeEventGroup ( eventGroup_t & eventGroup )

Program uncore PCIe monitoring event(s)

Parameters

eventGroup - events to program for the same run

◆ programServerUncoreLatencyMetrics()

PCM::ErrorCode pcm::PCM::programServerUncoreLatencyMetrics ( bool enable_pmm )

Programs uncore latency counters on microarchitectures codename SandyBridge-EP and later Xeon uarch.

Parameters

enable_pmm enables DDR/PMM. See possible profile values in pcm-latency.cpp example

Call this method before you start using the latency counter routines on microarchitecture codename SandyBridge-EP and later Xeon uarch

Warning: After this call the memory and QPI bandwidth counters on microarchitecture codename SandyBridge-EP and later Xeon uarch will not work.; Using this routines with other tools that program Performance Monitoring Units (PMUs) on CPUs is not recommended because PMU can not be shared. Tools that are known to program PMUs: Intel(r) VTune(tm), Intel(r) Performance Tuning Utility (PTU). This code may make VTune or PTU measurements invalid. VTune or PTU measurement may make measurement with this code invalid. Please enable either usage of these routines or VTune/PTU/etc.

◆ programServerUncorePowerMetrics()

PCM::ErrorCode pcm::PCM::programServerUncorePowerMetrics	(	int	mc_profile,
		int	pcu_profile,
		int *	freq_bands = `NULL`
	)

Programs uncore power/energy counters on microarchitectures codename SandyBridge-EP and later Xeon uarch.

Parameters

mc_profile	profile for integrated memory controller PMU. See possible profile values in pcm-power.cpp example
pcu_profile	profile for power control unit PMU. See possible profile values in pcm-power.cpp example
freq_bands	array of three integer values for core frequency band monitoring. See usage in pcm-power.cpp example

Call this method before you start using the power counter routines on microarchitecture codename SandyBridge-EP and later Xeon uarch

Warning: After this call the memory and QPI bandwidth counters on microarchitecture codename SandyBridge-EP and later Xeon uarch will not work.; Using this routines with other tools that program Performance Monitoring Units (PMUs) on CPUs is not recommended because PMU can not be shared. Tools that are known to program PMUs: Intel(r) VTune(tm), Intel(r) Performance Tuning Utility (PTU). This code may make VTune or PTU measurements invalid. VTune or PTU measurement may make measurement with this code invalid. Please enable either usage of these routines or VTune/PTU/etc.

◆ QOSMetricAvailable()

bool pcm::PCM::QOSMetricAvailable ( ) const

checks if QOS monitoring support present

Returns: true or false

References isSecureBoot().

Referenced by CoreLocalMemoryBWMetricAvailable(), CoreRemoteMemoryBWMetricAvailable(), and L3CacheOccupancyMetricAvailable().

◆ resetPMU()

void pcm::PCM::resetPMU ( )

Forces PMU reset.

If there is no chance to free up PMU from other applications you might try to call this method at your own risk.

Referenced by checkError().

◆ setupCustomCoreEventsForNuma()

void pcm::PCM::setupCustomCoreEventsForNuma ( PCM::ExtendedCustomCoreEventDescription & conf ) const

Setup ExtendedCustomCoreEventDescription object to read offcore (numa) counters for each processor type.

Parameters

conf	conf object to setup offcore MSR values

The documentation for this class was generated from the following files:

src/cpucounters.h
src/cpucounters.cpp

Classes

Public Types

Public Member Functions

Static Public Member Functions

Friends

Detailed Description

Member Enumeration Documentation

◆ ProgramMode

Member Function Documentation

◆ checkError()

◆ cleanup()

◆ CoreLocalMemoryBWMetricAvailable()

◆ CoreRemoteMemoryBWMetricAvailable()

◆ getAllCounterStates()

◆ getCoreCounterState()

◆ getCoreId()

◆ getCPUModel()

◆ getCPUStepping()

◆ getErrorMessage()

◆ getIIOCounterState()

◆ getIIOCounterStates()

◆ getInstance()

◆ getL3ScalingFactor()

◆ getMaxIPC()

◆ getMaxRMID()

◆ getMCChannels()

◆ getNominalFrequency()

◆ getNumCores()

◆ getNumOnlineCores()

◆ getNumOnlineSockets()

◆ getNumSockets()

◆ getPCIeCounterState()

◆ getQPILinkSpeed()

◆ getQPILinksPerSocket()

◆ getServerUncoreCounterState()

◆ getSMT()

◆ getSocketCounterState()

◆ getSocketId()

◆ getSystemCounterState()

◆ getSystemTopology()

◆ getThreadId()

◆ getThreadsPerCore()

◆ getTickCount()

◆ getTileId()

◆ getUArchCodename()

◆ getUncoreCounterStates()

◆ good()

◆ initWinRing0Lib()

◆ isCoreOnline()

◆ isSocketOnline()

◆ isSomeCoreOfflined()

◆ L3CacheOccupancyMetricAvailable()

◆ L3QOSMetricAvailable()

◆ program()

◆ programCbo()

◆ programCboRaw()

◆ programIIOCounters()

◆ programIRPCounters()

◆ programPCIeEventGroup()

◆ programServerUncoreLatencyMetrics()

◆ programServerUncorePowerMetrics()

◆ QOSMetricAvailable()

◆ resetPMU()

◆ setupCustomCoreEventsForNuma()